includes/Sanitizer.php

   1 <?php
   2 /**
   3  * HTML sanitizer for %MediaWiki.
   4  *
   5  * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * HTML sanitizer for MediaWiki
  29  * @ingroup Parser
  30  */
  31 class Sanitizer {
  32         /**
  33          * Regular expression to match various types of character references in
  34          * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  35          */
  36         const CHAR_REFS_REGEX =
  37                 '/&([A-Za-z0-9\x80-\xff]+);
  38                  |&\#([0-9]+);
  39                  |&\#[xX]([0-9A-Fa-f]+);
  40                  |(&)/x';
  41
  42         /**
  43          * Acceptable tag name charset from HTML5 parsing spec
  44          * http://www.w3.org/TR/html5/syntax.html#tag-open-state
  45          */
  46         const ELEMENT_BITS_REGEX = '!^(/?)([A-Za-z][^\t\n\v />\0]*+)([^>]*?)(/?>)([^<]*)$!';
  47
  48         /**
  49          * Blacklist for evil uris like javascript:
  50          * WARNING: DO NOT use this in any place that actually requires blacklisting
  51          * for security reasons. There are NUMEROUS[1] ways to bypass blacklisting, the
  52          * only way to be secure from javascript: uri based xss vectors is to whitelist
  53          * things that you know are safe and deny everything else.
  54          * [1]: http://ha.ckers.org/xss.html
  55          */
  56         const EVIL_URI_PATTERN = '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i';
  57         const XMLNS_ATTRIBUTE_PATTERN = "/^xmlns:[:A-Z_a-z-.0-9]+$/";
  58
  59         /**
  60          * List of all named character entities defined in HTML 4.01
  61          * http://www.w3.org/TR/html4/sgml/entities.html
  62          * As well as &apos; which is only defined starting in XHTML1.
  63          */
  64         private static $htmlEntities = array(
  65                 'Aacute'   => 193,
  66                 'aacute'   => 225,
  67                 'Acirc'    => 194,
  68                 'acirc'    => 226,
  69                 'acute'    => 180,
  70                 'AElig'    => 198,
  71                 'aelig'    => 230,
  72                 'Agrave'   => 192,
  73                 'agrave'   => 224,
  74                 'alefsym'  => 8501,
  75                 'Alpha'    => 913,
  76                 'alpha'    => 945,
  77                 'amp'      => 38,
  78                 'and'      => 8743,
  79                 'ang'      => 8736,
  80                 'apos'     => 39, // New in XHTML & HTML 5; avoid in output for compatibility with IE.
  81                 'Aring'    => 197,
  82                 'aring'    => 229,
  83                 'asymp'    => 8776,
  84                 'Atilde'   => 195,
  85                 'atilde'   => 227,
  86                 'Auml'     => 196,
  87                 'auml'     => 228,
  88                 'bdquo'    => 8222,
  89                 'Beta'     => 914,
  90                 'beta'     => 946,
  91                 'brvbar'   => 166,
  92                 'bull'     => 8226,
  93                 'cap'      => 8745,
  94                 'Ccedil'   => 199,
  95                 'ccedil'   => 231,
  96                 'cedil'    => 184,
  97                 'cent'     => 162,
  98                 'Chi'      => 935,
  99                 'chi'      => 967,
 100                 'circ'     => 710,
 101                 'clubs'    => 9827,
 102                 'cong'     => 8773,
 103                 'copy'     => 169,
 104                 'crarr'    => 8629,
 105                 'cup'      => 8746,
 106                 'curren'   => 164,
 107                 'dagger'   => 8224,
 108                 'Dagger'   => 8225,
 109                 'darr'     => 8595,
 110                 'dArr'     => 8659,
 111                 'deg'      => 176,
 112                 'Delta'    => 916,
 113                 'delta'    => 948,
 114                 'diams'    => 9830,
 115                 'divide'   => 247,
 116                 'Eacute'   => 201,
 117                 'eacute'   => 233,
 118                 'Ecirc'    => 202,
 119                 'ecirc'    => 234,
 120                 'Egrave'   => 200,
 121                 'egrave'   => 232,
 122                 'empty'    => 8709,
 123                 'emsp'     => 8195,
 124                 'ensp'     => 8194,
 125                 'Epsilon'  => 917,
 126                 'epsilon'  => 949,
 127                 'equiv'    => 8801,
 128                 'Eta'      => 919,
 129                 'eta'      => 951,
 130                 'ETH'      => 208,
 131                 'eth'      => 240,
 132                 'Euml'     => 203,
 133                 'euml'     => 235,
 134                 'euro'     => 8364,
 135                 'exist'    => 8707,
 136                 'fnof'     => 402,
 137                 'forall'   => 8704,
 138                 'frac12'   => 189,
 139                 'frac14'   => 188,
 140                 'frac34'   => 190,
 141                 'frasl'    => 8260,
 142                 'Gamma'    => 915,
 143                 'gamma'    => 947,
 144                 'ge'       => 8805,
 145                 'gt'       => 62,
 146                 'harr'     => 8596,
 147                 'hArr'     => 8660,
 148                 'hearts'   => 9829,
 149                 'hellip'   => 8230,
 150                 'Iacute'   => 205,
 151                 'iacute'   => 237,
 152                 'Icirc'    => 206,
 153                 'icirc'    => 238,
 154                 'iexcl'    => 161,
 155                 'Igrave'   => 204,
 156                 'igrave'   => 236,
 157                 'image'    => 8465,
 158                 'infin'    => 8734,
 159                 'int'      => 8747,
 160                 'Iota'     => 921,
 161                 'iota'     => 953,
 162                 'iquest'   => 191,
 163                 'isin'     => 8712,
 164                 'Iuml'     => 207,
 165                 'iuml'     => 239,
 166                 'Kappa'    => 922,
 167                 'kappa'    => 954,
 168                 'Lambda'   => 923,
 169                 'lambda'   => 955,
 170                 'lang'     => 9001,
 171                 'laquo'    => 171,
 172                 'larr'     => 8592,
 173                 'lArr'     => 8656,
 174                 'lceil'    => 8968,
 175                 'ldquo'    => 8220,
 176                 'le'       => 8804,
 177                 'lfloor'   => 8970,
 178                 'lowast'   => 8727,
 179                 'loz'      => 9674,
 180                 'lrm'      => 8206,
 181                 'lsaquo'   => 8249,
 182                 'lsquo'    => 8216,
 183                 'lt'       => 60,
 184                 'macr'     => 175,
 185                 'mdash'    => 8212,
 186                 'micro'    => 181,
 187                 'middot'   => 183,
 188                 'minus'    => 8722,
 189                 'Mu'       => 924,
 190                 'mu'       => 956,
 191                 'nabla'    => 8711,
 192                 'nbsp'     => 160,
 193                 'ndash'    => 8211,
 194                 'ne'       => 8800,
 195                 'ni'       => 8715,
 196                 'not'      => 172,
 197                 'notin'    => 8713,
 198                 'nsub'     => 8836,
 199                 'Ntilde'   => 209,
 200                 'ntilde'   => 241,
 201                 'Nu'       => 925,
 202                 'nu'       => 957,
 203                 'Oacute'   => 211,
 204                 'oacute'   => 243,
 205                 'Ocirc'    => 212,
 206                 'ocirc'    => 244,
 207                 'OElig'    => 338,
 208                 'oelig'    => 339,
 209                 'Ograve'   => 210,
 210                 'ograve'   => 242,
 211                 'oline'    => 8254,
 212                 'Omega'    => 937,
 213                 'omega'    => 969,
 214                 'Omicron'  => 927,
 215                 'omicron'  => 959,
 216                 'oplus'    => 8853,
 217                 'or'       => 8744,
 218                 'ordf'     => 170,
 219                 'ordm'     => 186,
 220                 'Oslash'   => 216,
 221                 'oslash'   => 248,
 222                 'Otilde'   => 213,
 223                 'otilde'   => 245,
 224                 'otimes'   => 8855,
 225                 'Ouml'     => 214,
 226                 'ouml'     => 246,
 227                 'para'     => 182,
 228                 'part'     => 8706,
 229                 'permil'   => 8240,
 230                 'perp'     => 8869,
 231                 'Phi'      => 934,
 232                 'phi'      => 966,
 233                 'Pi'       => 928,
 234                 'pi'       => 960,
 235                 'piv'      => 982,
 236                 'plusmn'   => 177,
 237                 'pound'    => 163,
 238                 'prime'    => 8242,
 239                 'Prime'    => 8243,
 240                 'prod'     => 8719,
 241                 'prop'     => 8733,
 242                 'Psi'      => 936,
 243                 'psi'      => 968,
 244                 'quot'     => 34,
 245                 'radic'    => 8730,
 246                 'rang'     => 9002,
 247                 'raquo'    => 187,
 248                 'rarr'     => 8594,
 249                 'rArr'     => 8658,
 250                 'rceil'    => 8969,
 251                 'rdquo'    => 8221,
 252                 'real'     => 8476,
 253                 'reg'      => 174,
 254                 'rfloor'   => 8971,
 255                 'Rho'      => 929,
 256                 'rho'      => 961,
 257                 'rlm'      => 8207,
 258                 'rsaquo'   => 8250,
 259                 'rsquo'    => 8217,
 260                 'sbquo'    => 8218,
 261                 'Scaron'   => 352,
 262                 'scaron'   => 353,
 263                 'sdot'     => 8901,
 264                 'sect'     => 167,
 265                 'shy'      => 173,
 266                 'Sigma'    => 931,
 267                 'sigma'    => 963,
 268                 'sigmaf'   => 962,
 269                 'sim'      => 8764,
 270                 'spades'   => 9824,
 271                 'sub'      => 8834,
 272                 'sube'     => 8838,
 273                 'sum'      => 8721,
 274                 'sup'      => 8835,
 275                 'sup1'     => 185,
 276                 'sup2'     => 178,
 277                 'sup3'     => 179,
 278                 'supe'     => 8839,
 279                 'szlig'    => 223,
 280                 'Tau'      => 932,
 281                 'tau'      => 964,
 282                 'there4'   => 8756,
 283                 'Theta'    => 920,
 284                 'theta'    => 952,
 285                 'thetasym' => 977,
 286                 'thinsp'   => 8201,
 287                 'THORN'    => 222,
 288                 'thorn'    => 254,
 289                 'tilde'    => 732,
 290                 'times'    => 215,
 291                 'trade'    => 8482,
 292                 'Uacute'   => 218,
 293                 'uacute'   => 250,
 294                 'uarr'     => 8593,
 295                 'uArr'     => 8657,
 296                 'Ucirc'    => 219,
 297                 'ucirc'    => 251,
 298                 'Ugrave'   => 217,
 299                 'ugrave'   => 249,
 300                 'uml'      => 168,
 301                 'upsih'    => 978,
 302                 'Upsilon'  => 933,
 303                 'upsilon'  => 965,
 304                 'Uuml'     => 220,
 305                 'uuml'     => 252,
 306                 'weierp'   => 8472,
 307                 'Xi'       => 926,
 308                 'xi'       => 958,
 309                 'Yacute'   => 221,
 310                 'yacute'   => 253,
 311                 'yen'      => 165,
 312                 'Yuml'     => 376,
 313                 'yuml'     => 255,
 314                 'Zeta'     => 918,
 315                 'zeta'     => 950,
 316                 'zwj'      => 8205,
 317                 'zwnj'     => 8204
 318         );
 319
 320         /**
 321          * Character entity aliases accepted by MediaWiki
 322          */
 323         private static $htmlEntityAliases = array(
 324                 'רלמ' => 'rlm',
 325                 'رلم' => 'rlm',
 326         );
 327
 328         /**
 329          * Lazy-initialised attributes regex, see getAttribsRegex()
 330          */
 331         private static $attribsRegex;
 332
 333         /**
 334          * Regular expression to match HTML/XML attribute pairs within a tag.
 335          * Allows some... latitude.
 336          * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
 337          * @return string
 338          */
 339         static function getAttribsRegex() {
 340                 if ( self::$attribsRegex === null ) {
 341                         $attribFirst = '[:A-Z_a-z0-9]';
 342                         $attrib = '[:A-Z_a-z-.0-9]';
 343                         $space = '[\x09\x0a\x0d\x20]';
 344                         self::$attribsRegex =
 345                                 "/(?:^|$space)({$attribFirst}{$attrib}*)
 346                                   ($space*=$space*
 347                                         (?:
 348                                          # The attribute value: quoted or alone
 349                                           \"([^<\"]*)\"
 350                                          | '([^<']*)'
 351                                          |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
 352                                         )
 353                                 )?(?=$space|\$)/sx";
 354                 }
 355                 return self::$attribsRegex;
 356         }
 357
 358         /**
 359          * Return the various lists of recognized tags
 360          * @param array $extratags For any extra tags to include
 361          * @param array $removetags For any tags (default or extra) to exclude
 362          * @return array
 363          */
 364         public static function getRecognizedTagData( $extratags = array(), $removetags = array() ) {
 365                 global $wgAllowMicrodataAttributes, $wgAllowImageTag;
 366
 367                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 368                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 369
 370                 // Base our staticInitialised variable off of the global config state so that if the globals
 371                 // are changed (like in the screwed up test system) we will re-initialise the settings.
 372                 $globalContext = implode( '-', compact( 'wgAllowMicrodataAttributes', 'wgAllowImageTag' ) );
 373                 if ( !$staticInitialised || $staticInitialised != $globalContext ) {
 374                         $htmlpairsStatic = array( # Tags that must be closed
 375                                 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 376                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 377                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 378                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 379                                 'ruby', 'rb', 'rp', 'rt', 'rtc', 'p', 'span', 'abbr', 'dfn',
 380                                 'kbd', 'samp', 'data', 'time', 'mark'
 381                         );
 382                         $htmlsingle = array(
 383                                 'br', 'wbr', 'hr', 'li', 'dt', 'dd'
 384                         );
 385                         $htmlsingleonly = array( # Elements that cannot have close tags
 386                                 'br', 'wbr', 'hr'
 387                         );
 388                         if ( $wgAllowMicrodataAttributes ) {
 389                                 $htmlsingle[] = $htmlsingleonly[] = 'meta';
 390                                 $htmlsingle[] = $htmlsingleonly[] = 'link';
 391                         }
 392                         $htmlnest = array( # Tags that can be nested--??
 393                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 394                                 'li', 'dl', 'dt', 'dd', 'font', 'big', 'small', 'sub', 'sup', 'span',
 395                                 'var', 'kbd', 'samp', 'em', 'strong', 'q', 'ruby', 'bdo'
 396                         );
 397                         $tabletags = array( # Can only appear inside table, we will close them
 398                                 'td', 'th', 'tr',
 399                         );
 400                         $htmllist = array( # Tags used by list
 401                                 'ul', 'ol',
 402                         );
 403                         $listtags = array( # Tags that can appear in a list
 404                                 'li',
 405                         );
 406
 407                         if ( $wgAllowImageTag ) {
 408                                 $htmlsingle[] = 'img';
 409                                 $htmlsingleonly[] = 'img';
 410                         }
 411
 412                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 413                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 414
 415                         # Convert them all to hashtables for faster lookup
 416                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 417                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 418                         foreach ( $vars as $var ) {
 419                                 $$var = array_flip( $$var );
 420                         }
 421                         $staticInitialised = $globalContext;
 422                 }
 423
 424                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 425                 $extratags = array_flip( $extratags );
 426                 $removetags = array_flip( $removetags );
 427                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 428                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ), $removetags );
 429
 430                 return array(
 431                         'htmlpairs' => $htmlpairs,
 432                         'htmlsingle' => $htmlsingle,
 433                         'htmlsingleonly' => $htmlsingleonly,
 434                         'htmlnest' => $htmlnest,
 435                         'tabletags' => $tabletags,
 436                         'htmllist' => $htmllist,
 437                         'listtags' => $listtags,
 438                         'htmlsingleallowed' => $htmlsingleallowed,
 439                         'htmlelements' => $htmlelements,
 440                 );
 441         }
 442
 443         /**
 444          * Cleans up HTML, removes dangerous tags and attributes, and
 445          * removes HTML comments
 446          * @param string $text
 447          * @param callable $processCallback Callback to do any variable or parameter
 448          *   replacements in HTML attribute values
 449          * @param array|bool $args Arguments for the processing callback
 450          * @param array $extratags For any extra tags to include
 451          * @param array $removetags For any tags (default or extra) to exclude
 452          * @return string
 453          */
 454         public static function removeHTMLtags( $text, $processCallback = null,
 455                 $args = array(), $extratags = array(), $removetags = array()
 456         ) {
 457                 global $wgUseTidy;
 458
 459                 extract( self::getRecognizedTagData( $extratags, $removetags ) );
 460
 461                 # Remove HTML comments
 462                 $text = Sanitizer::removeHTMLcomments( $text );
 463                 $bits = explode( '<', $text );
 464                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 465                 if ( !$wgUseTidy ) {
 466                         $tagstack = $tablestack = array();
 467                         foreach ( $bits as $x ) {
 468                                 $regs = array();
 469                                 # $slash: Does the current element start with a '/'?
 470                                 # $t: Current element name
 471                                 # $params: String between element name and >
 472                                 # $brace: Ending '>' or '/>'
 473                                 # $rest: Everything until the next element of $bits
 474                                 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
 475                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 476                                 } else {
 477                                         $slash = $t = $params = $brace = $rest = null;
 478                                 }
 479
 480                                 $badtag = false;
 481                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 482                                         # Check our stack
 483                                         if ( $slash && isset( $htmlsingleonly[$t] ) ) {
 484                                                 $badtag = true;
 485                                         } elseif ( $slash ) {
 486                                                 # Closing a tag... is it the one we just opened?
 487                                                 MediaWiki\suppressWarnings();
 488                                                 $ot = array_pop( $tagstack );
 489                                                 MediaWiki\restoreWarnings();
 490
 491                                                 if ( $ot != $t ) {
 492                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 493                                                                 # Pop all elements with an optional close tag
 494                                                                 # and see if we find a match below them
 495                                                                 $optstack = array();
 496                                                                 array_push( $optstack, $ot );
 497                                                                 MediaWiki\suppressWarnings();
 498                                                                 $ot = array_pop( $tagstack );
 499                                                                 MediaWiki\restoreWarnings();
 500                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 501                                                                         array_push( $optstack, $ot );
 502                                                                         MediaWiki\suppressWarnings();
 503                                                                         $ot = array_pop( $tagstack );
 504                                                                         MediaWiki\restoreWarnings();
 505                                                                 }
 506                                                                 if ( $t != $ot ) {
 507                                                                         # No match. Push the optional elements back again
 508                                                                         $badtag = true;
 509                                                                         MediaWiki\suppressWarnings();
 510                                                                         $ot = array_pop( $optstack );
 511                                                                         MediaWiki\restoreWarnings();
 512                                                                         while ( $ot ) {
 513                                                                                 array_push( $tagstack, $ot );
 514                                                                                 MediaWiki\suppressWarnings();
 515                                                                                 $ot = array_pop( $optstack );
 516                                                                                 MediaWiki\restoreWarnings();
 517                                                                         }
 518                                                                 }
 519                                                         } else {
 520                                                                 MediaWiki\suppressWarnings();
 521                                                                 array_push( $tagstack, $ot );
 522                                                                 MediaWiki\restoreWarnings();
 523
 524                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 525                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 526                                                                         $badtag = true;
 527                                                                 }
 528                                                         }
 529                                                 } else {
 530                                                         if ( $t == 'table' ) {
 531                                                                 $tagstack = array_pop( $tablestack );
 532                                                         }
 533                                                 }
 534                                                 $newparams = '';
 535                                         } else {
 536                                                 # Keep track for later
 537                                                 if ( isset( $tabletags[$t] ) && !in_array( 'table', $tagstack ) ) {
 538                                                         $badtag = true;
 539                                                 } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
 540                                                         $badtag = true;
 541                                                 # Is it a self closed htmlpair ? (bug 5487)
 542                                                 } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
 543                                                         $badtag = true;
 544                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 545                                                         # Hack to force empty tag for unclosable elements
 546                                                         $brace = '/>';
 547                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 548                                                         # Hack to not close $htmlsingle tags
 549                                                         $brace = null;
 550                                                         # Still need to push this optionally-closed tag to
 551                                                         # the tag stack so that we can match end tags
 552                                                         # instead of marking them as bad.
 553                                                         array_push( $tagstack, $t );
 554                                                 } elseif ( isset( $tabletags[$t] ) && in_array( $t, $tagstack ) ) {
 555                                                         // New table tag but forgot to close the previous one
 556                                                         $text .= "</$t>";
 557                                                 } else {
 558                                                         if ( $t == 'table' ) {
 559                                                                 array_push( $tablestack, $tagstack );
 560                                                                 $tagstack = array();
 561                                                         }
 562                                                         array_push( $tagstack, $t );
 563                                                 }
 564
 565                                                 # Replace any variables or template parameters with
 566                                                 # plaintext results.
 567                                                 if ( is_callable( $processCallback ) ) {
 568                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 569                                                 }
 570
 571                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 572                                                         $badtag = true;
 573                                                 }
 574
 575                                                 # Strip non-approved attributes from the tag
 576                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 577                                         }
 578                                         if ( !$badtag ) {
 579                                                 $rest = str_replace( '>', '&gt;', $rest );
 580                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 581                                                 $text .= "<$slash$t$newparams$close>$rest";
 582                                                 continue;
 583                                         }
 584                                 }
 585                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 586                         }
 587                         # Close off any remaining tags
 588                         while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
 589                                 $text .= "</$t>\n";
 590                                 if ( $t == 'table' ) {
 591                                         $tagstack = array_pop( $tablestack );
 592                                 }
 593                         }
 594                 } else {
 595                         # this might be possible using tidy itself
 596                         foreach ( $bits as $x ) {
 597                                 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
 598                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 599
 600                                         $badtag = false;
 601                                         if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 602                                                 if ( is_callable( $processCallback ) ) {
 603                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 604                                                 }
 605
 606                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 607                                                         $badtag = true;
 608                                                 }
 609
 610                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 611                                                 if ( !$badtag ) {
 612                                                         $rest = str_replace( '>', '&gt;', $rest );
 613                                                         $text .= "<$slash$t$newparams$brace$rest";
 614                                                         continue;
 615                                                 }
 616                                         }
 617                                 }
 618                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 619                         }
 620                 }
 621                 return $text;
 622         }
 623
 624         /**
 625          * Remove '<!--', '-->', and everything between.
 626          * To avoid leaving blank lines, when a comment is both preceded
 627          * and followed by a newline (ignoring spaces), trim leading and
 628          * trailing spaces and one of the newlines.
 629          *
 630          * @param string $text
 631          * @return string
 632          */
 633         public static function removeHTMLcomments( $text ) {
 634                 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
 635                         $end = strpos( $text, '-->', $start + 4 );
 636                         if ( $end === false ) {
 637                                 # Unterminated comment; bail out
 638                                 break;
 639                         }
 640
 641                         $end += 3;
 642
 643                         # Trim space and newline if the comment is both
 644                         # preceded and followed by a newline
 645                         $spaceStart = max( $start - 1, 0 );
 646                         $spaceLen = $end - $spaceStart;
 647                         while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
 648                                 $spaceStart--;
 649                                 $spaceLen++;
 650                         }
 651                         while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
 652                                 $spaceLen++;
 653                         }
 654                         if ( substr( $text, $spaceStart, 1 ) === "\n"
 655                                 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
 656                                 # Remove the comment, leading and trailing
 657                                 # spaces, and leave only one newline.
 658                                 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
 659                         } else {
 660                                 # Remove just the comment.
 661                                 $text = substr_replace( $text, '', $start, $end - $start );
 662                         }
 663                 }
 664                 return $text;
 665         }
 666
 667         /**
 668          * Takes attribute names and values for a tag and the tag name and
 669          * validates that the tag is allowed to be present.
 670          * This DOES NOT validate the attributes, nor does it validate the
 671          * tags themselves. This method only handles the special circumstances
 672          * where we may want to allow a tag within content but ONLY when it has
 673          * specific attributes set.
 674          *
 675          * @param string $params
 676          * @param string $element
 677          * @return bool
 678          */
 679         static function validateTag( $params, $element ) {
 680                 $params = Sanitizer::decodeTagAttributes( $params );
 681
 682                 if ( $element == 'meta' || $element == 'link' ) {
 683                         if ( !isset( $params['itemprop'] ) ) {
 684                                 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
 685                                 return false;
 686                         }
 687                         if ( $element == 'meta' && !isset( $params['content'] ) ) {
 688                                 // <meta> must have a content="" for the itemprop
 689                                 return false;
 690                         }
 691                         if ( $element == 'link' && !isset( $params['href'] ) ) {
 692                                 // <link> must have an associated href=""
 693                                 return false;
 694                         }
 695                 }
 696
 697                 return true;
 698         }
 699
 700         /**
 701          * Take an array of attribute names and values and normalize or discard
 702          * illegal values for the given element type.
 703          *
 704          * - Discards attributes not on a whitelist for the given element
 705          * - Unsafe style attributes are discarded
 706          * - Invalid id attributes are re-encoded
 707          *
 708          * @param array $attribs
 709          * @param string $element
 710          * @return array
 711          *
 712          * @todo Check for legal values where the DTD limits things.
 713          * @todo Check for unique id attribute :P
 714          */
 715         static function validateTagAttributes( $attribs, $element ) {
 716                 return Sanitizer::validateAttributes( $attribs,
 717                         Sanitizer::attributeWhitelist( $element ) );
 718         }
 719
 720         /**
 721          * Take an array of attribute names and values and normalize or discard
 722          * illegal values for the given whitelist.
 723          *
 724          * - Discards attributes not the given whitelist
 725          * - Unsafe style attributes are discarded
 726          * - Invalid id attributes are re-encoded
 727          *
 728          * @param array $attribs
 729          * @param array $whitelist List of allowed attribute names
 730          * @return array
 731          *
 732          * @todo Check for legal values where the DTD limits things.
 733          * @todo Check for unique id attribute :P
 734          */
 735         static function validateAttributes( $attribs, $whitelist ) {
 736                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 737
 738                 $whitelist = array_flip( $whitelist );
 739                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 740
 741                 $out = array();
 742                 foreach ( $attribs as $attribute => $value ) {
 743                         #allow XML namespace declaration if RDFa is enabled
 744                         if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
 745                                 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 746                                         $out[$attribute] = $value;
 747                                 }
 748
 749                                 continue;
 750                         }
 751
 752                         # Allow any attribute beginning with "data-"
 753                         if ( !preg_match( '/^data-(?!ooui)/i', $attribute ) && !isset( $whitelist[$attribute] ) ) {
 754                                 continue;
 755                         }
 756
 757                         # Strip javascript "expression" from stylesheets.
 758                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 759                         if ( $attribute == 'style' ) {
 760                                 $value = Sanitizer::checkCss( $value );
 761                         }
 762
 763                         if ( $attribute === 'id' ) {
 764                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 765                         }
 766
 767                         # WAI-ARIA
 768                         # http://www.w3.org/TR/wai-aria/
 769                         # http://www.whatwg.org/html/elements.html#wai-aria
 770                         # For now we only support role="presentation" until we work out what roles should be
 771                         # usable by content and we ensure that our code explicitly rejects patterns that
 772                         # violate HTML5's ARIA restrictions.
 773                         if ( $attribute === 'role' && $value !== 'presentation' ) {
 774                                 continue;
 775                         }
 776
 777                         // RDFa and microdata properties allow URLs, URIs and/or CURIs.
 778                         // Check them for sanity.
 779                         if ( $attribute === 'rel' || $attribute === 'rev'
 780                                 # RDFa
 781                                 || $attribute === 'about' || $attribute === 'property'
 782                                 || $attribute === 'resource' || $attribute === 'datatype'
 783                                 || $attribute === 'typeof'
 784                                 # HTML5 microdata
 785                                 || $attribute === 'itemid' || $attribute === 'itemprop'
 786                                 || $attribute === 'itemref' || $attribute === 'itemscope'
 787                                 || $attribute === 'itemtype'
 788                         ) {
 789                                 //Paranoia. Allow "simple" values but suppress javascript
 790                                 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 791                                         continue;
 792                                 }
 793                         }
 794
 795                         # NOTE: even though elements using href/src are not allowed directly, supply
 796                         #       validation code that can be used by tag hook handlers, etc
 797                         if ( $attribute === 'href' || $attribute === 'src' ) {
 798                                 if ( !preg_match( $hrefExp, $value ) ) {
 799                                         continue; //drop any href or src attributes not using an allowed protocol.
 800                                         // NOTE: this also drops all relative URLs
 801                                 }
 802                         }
 803
 804                         // If this attribute was previously set, override it.
 805                         // Output should only have one attribute of each name.
 806                         $out[$attribute] = $value;
 807                 }
 808
 809                 if ( $wgAllowMicrodataAttributes ) {
 810                         # itemtype, itemid, itemref don't make sense without itemscope
 811                         if ( !array_key_exists( 'itemscope', $out ) ) {
 812                                 unset( $out['itemtype'] );
 813                                 unset( $out['itemid'] );
 814                                 unset( $out['itemref'] );
 815                         }
 816                         # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
 817                 }
 818                 return $out;
 819         }
 820
 821         /**
 822          * Merge two sets of HTML attributes.  Conflicting items in the second set
 823          * will override those in the first, except for 'class' attributes which
 824          * will be combined (if they're both strings).
 825          *
 826          * @todo implement merging for other attributes such as style
 827          * @param array $a
 828          * @param array $b
 829          * @return array
 830          */
 831         static function mergeAttributes( $a, $b ) {
 832                 $out = array_merge( $a, $b );
 833                 if ( isset( $a['class'] ) && isset( $b['class'] )
 834                         && is_string( $a['class'] ) && is_string( $b['class'] )
 835                         && $a['class'] !== $b['class']
 836                 ) {
 837                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 838                                 -1, PREG_SPLIT_NO_EMPTY );
 839                         $out['class'] = implode( ' ', array_unique( $classes ) );
 840                 }
 841                 return $out;
 842         }
 843
 844         /**
 845          * Normalize CSS into a format we can easily search for hostile input
 846          *  - decode character references
 847          *  - decode escape sequences
 848          *  - convert characters that IE6 interprets into ascii
 849          *  - remove comments, unless the entire value is one single comment
 850          * @param string $value the css string
 851          * @return string normalized css
 852          */
 853         public static function normalizeCss( $value ) {
 854
 855                 // Decode character references like &#123;
 856                 $value = Sanitizer::decodeCharReferences( $value );
 857
 858                 // Decode escape sequences and line continuation
 859                 // See the grammar in the CSS 2 spec, appendix D.
 860                 // This has to be done AFTER decoding character references.
 861                 // This means it isn't possible for this function to return
 862                 // unsanitized escape sequences. It is possible to manufacture
 863                 // input that contains character references that decode to
 864                 // escape sequences that decode to character references, but
 865                 // it's OK for the return value to contain character references
 866                 // because the caller is supposed to escape those anyway.
 867                 static $decodeRegex;
 868                 if ( !$decodeRegex ) {
 869                         $space = '[\\x20\\t\\r\\n\\f]';
 870                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 871                         $backslash = '\\\\';
 872                         $decodeRegex = "/ $backslash
 873                                 (?:
 874                                         ($nl) |  # 1. Line continuation
 875                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 876                                         (.) | # 3. backslash cancelling special meaning
 877                                         () | # 4. backslash at end of string
 878                                 )/xu";
 879                 }
 880                 $value = preg_replace_callback( $decodeRegex,
 881                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 882
 883                 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
 884                 $value = preg_replace_callback(
 885                         '/[！-［］-ｚ]/u', // U+FF01 to U+FF5A, excluding U+FF3C (bug 58088)
 886                         function ( $matches ) {
 887                                 $cp = UtfNormal\Utils::utf8ToCodepoint( $matches[0] );
 888                                 if ( $cp === false ) {
 889                                         return '';
 890                                 }
 891                                 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
 892                         },
 893                         $value
 894                 );
 895
 896                 // Convert more characters IE6 might treat as ascii
 897                 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
 898                 $value = str_replace(
 899                         array( 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ),
 900                         array( 'r', 'n', 'n', 'l', 'i', '(', '(' ),
 901                         $value
 902                 );
 903
 904                 // Let the value through if it's nothing but a single comment, to
 905                 // allow other functions which may reject it to pass some error
 906                 // message through.
 907                 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
 908                         // Remove any comments; IE gets token splitting wrong
 909                         // This must be done AFTER decoding character references and
 910                         // escape sequences, because those steps can introduce comments
 911                         // This step cannot introduce character references or escape
 912                         // sequences, because it replaces comments with spaces rather
 913                         // than removing them completely.
 914                         $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 915
 916                         // Remove anything after a comment-start token, to guard against
 917                         // incorrect client implementations.
 918                         $commentPos = strpos( $value, '/*' );
 919                         if ( $commentPos !== false ) {
 920                                 $value = substr( $value, 0, $commentPos );
 921                         }
 922                 }
 923
 924                 // S followed by repeat, iteration, or prolonged sound marks,
 925                 // which IE will treat as "ss"
 926                 $value = preg_replace(
 927                         '/s(?:
 928                                 \xE3\x80\xB1 | # U+3031
 929                                 \xE3\x82\x9D | # U+309D
 930                                 \xE3\x83\xBC | # U+30FC
 931                                 \xE3\x83\xBD | # U+30FD
 932                                 \xEF\xB9\xBC | # U+FE7C
 933                                 \xEF\xB9\xBD | # U+FE7D
 934                                 \xEF\xBD\xB0   # U+FF70
 935                         )/ix',
 936                         'ss',
 937                         $value
 938                 );
 939
 940                 return $value;
 941         }
 942
 943
 944         /**
 945          * Pick apart some CSS and check it for forbidden or unsafe structures.
 946          * Returns a sanitized string. This sanitized string will have
 947          * character references and escape sequences decoded and comments
 948          * stripped (unless it is itself one valid comment, in which case the value
 949          * will be passed through). If the input is just too evil, only a comment
 950          * complaining about evilness will be returned.
 951          *
 952          * Currently URL references, 'expression', 'tps' are forbidden.
 953          *
 954          * NOTE: Despite the fact that character references are decoded, the
 955          * returned string may contain character references given certain
 956          * clever input strings. These character references must
 957          * be escaped before the return value is embedded in HTML.
 958          *
 959          * @param string $value
 960          * @return string
 961          */
 962         static function checkCss( $value ) {
 963                 $value = self::normalizeCss( $value );
 964
 965                 // Reject problematic keywords and control characters
 966                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
 967                         return '/* invalid control char */';
 968                 } elseif ( preg_match(
 969                         '! expression
 970                                 | filter\s*:
 971                                 | accelerator\s*:
 972                                 | -o-link\s*:
 973                                 | -o-link-source\s*:
 974                                 | -o-replace\s*:
 975                                 | url\s*\(
 976                                 | image\s*\(
 977                                 | image-set\s*\(
 978                         !ix', $value ) ) {
 979                         return '/* insecure input */';
 980                 }
 981                 return $value;
 982         }
 983
 984         /**
 985          * @param array $matches
 986          * @return string
 987          */
 988         static function cssDecodeCallback( $matches ) {
 989                 if ( $matches[1] !== '' ) {
 990                         // Line continuation
 991                         return '';
 992                 } elseif ( $matches[2] !== '' ) {
 993                         $char = UtfNormal\Utils::codepointToUtf8( hexdec( $matches[2] ) );
 994                 } elseif ( $matches[3] !== '' ) {
 995                         $char = $matches[3];
 996                 } else {
 997                         $char = '\\';
 998                 }
 999                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
1000                         // These characters need to be escaped in strings
1001                         // Clean up the escape sequence to avoid parsing errors by clients
1002                         return '\\' . dechex( ord( $char ) ) . ' ';
1003                 } else {
1004                         // Decode unnecessary escape
1005                         return $char;
1006                 }
1007         }
1008
1009         /**
1010          * Take a tag soup fragment listing an HTML element's attributes
1011          * and normalize it to well-formed XML, discarding unwanted attributes.
1012          * Output is safe for further wikitext processing, with escaping of
1013          * values that could trigger problems.
1014          *
1015          * - Normalizes attribute names to lowercase
1016          * - Discards attributes not on a whitelist for the given element
1017          * - Turns broken or invalid entities into plaintext
1018          * - Double-quotes all attribute values
1019          * - Attributes without values are given the name as attribute
1020          * - Double attributes are discarded
1021          * - Unsafe style attributes are discarded
1022          * - Prepends space if there are attributes.
1023          *
1024          * @param string $text
1025          * @param string $element
1026          * @return string
1027          */
1028         static function fixTagAttributes( $text, $element ) {
1029                 if ( trim( $text ) == '' ) {
1030                         return '';
1031                 }
1032
1033                 $decoded = Sanitizer::decodeTagAttributes( $text );
1034                 $stripped = Sanitizer::validateTagAttributes( $decoded, $element );
1035
1036                 return Sanitizer::safeEncodeTagAttributes( $stripped );
1037         }
1038
1039         /**
1040          * Encode an attribute value for HTML output.
1041          * @param string $text
1042          * @return string HTML-encoded text fragment
1043          */
1044         static function encodeAttribute( $text ) {
1045                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
1046
1047                 // Whitespace is normalized during attribute decoding,
1048                 // so if we've been passed non-spaces we must encode them
1049                 // ahead of time or they won't be preserved.
1050                 $encValue = strtr( $encValue, array(
1051                         "\n" => '&#10;',
1052                         "\r" => '&#13;',
1053                         "\t" => '&#9;',
1054                 ) );
1055
1056                 return $encValue;
1057         }
1058
1059         /**
1060          * Encode an attribute value for HTML tags, with extra armoring
1061          * against further wiki processing.
1062          * @param string $text
1063          * @return string HTML-encoded text fragment
1064          */
1065         static function safeEncodeAttribute( $text ) {
1066                 $encValue = Sanitizer::encodeAttribute( $text );
1067
1068                 # Templates and links may be expanded in later parsing,
1069                 # creating invalid or dangerous output. Suppress this.
1070                 $encValue = strtr( $encValue, array(
1071                         '<'    => '&lt;',   // This should never happen,
1072                         '>'    => '&gt;',   // we've received invalid input
1073                         '"'    => '&quot;', // which should have been escaped.
1074                         '{'    => '&#123;',
1075                         '['    => '&#91;',
1076                         "''"   => '&#39;&#39;',
1077                         'ISBN' => '&#73;SBN',
1078                         'RFC'  => '&#82;FC',
1079                         'PMID' => '&#80;MID',
1080                         '|'    => '&#124;',
1081                         '__'   => '&#95;_',
1082                 ) );
1083
1084                 # Stupid hack
1085                 $encValue = preg_replace_callback(
1086                         '/((?i)' . wfUrlProtocols() . ')/',
1087                         array( 'Sanitizer', 'armorLinksCallback' ),
1088                         $encValue );
1089                 return $encValue;
1090         }
1091
1092         /**
1093          * Given a value, escape it so that it can be used in an id attribute and
1094          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
1095          * true, allowing anything but ASCII whitespace.  Otherwise it will use
1096          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1097          * escaped with lots of dots.
1098          *
1099          * To ensure we don't have to bother escaping anything, we also strip ', ",
1100          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
1101          * We also strip # because it upsets IE, and % because it could be
1102          * ambiguous if it's part of something that looks like a percent escape
1103          * (which don't work reliably in fragments cross-browser).
1104          *
1105          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
1106          *   in the id and name attributes
1107          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1108          *   the id attribute
1109          * @see http://www.whatwg.org/html/elements.html#the-id-attribute
1110          *   HTML5 definition of id attribute
1111          *
1112          * @param string $id Id to escape
1113          * @param string|array $options String or array of strings (default is array()):
1114          *   'noninitial': This is a non-initial fragment of an id, not a full id,
1115          *       so don't pay attention if the first character isn't valid at the
1116          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
1117          *       false.
1118          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
1119          *       if $wgExperimentalHtmlIds is used, so we can generate extra
1120          *       anchors and links won't break.
1121          * @return string
1122          */
1123         static function escapeId( $id, $options = array() ) {
1124                 global $wgExperimentalHtmlIds;
1125                 $options = (array)$options;
1126
1127                 $id = Sanitizer::decodeCharReferences( $id );
1128
1129                 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1130                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1131                         $id = trim( $id, '_' );
1132                         if ( $id === '' ) {
1133                                 // Must have been all whitespace to start with.
1134                                 return '_';
1135                         } else {
1136                                 return $id;
1137                         }
1138                 }
1139
1140                 // HTML4-style escaping
1141                 static $replace = array(
1142                         '%3A' => ':',
1143                         '%' => '.'
1144                 );
1145
1146                 $id = urlencode( strtr( $id, ' ', '_' ) );
1147                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
1148
1149                 if ( !preg_match( '/^[a-zA-Z]/', $id ) && !in_array( 'noninitial', $options ) ) {
1150                         // Initial character must be a letter!
1151                         $id = "x$id";
1152                 }
1153                 return $id;
1154         }
1155
1156         /**
1157          * Given a value, escape it so that it can be used as a CSS class and
1158          * return it.
1159          *
1160          * @todo For extra validity, input should be validated UTF-8.
1161          *
1162          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1163          *
1164          * @param string $class
1165          * @return string
1166          */
1167         static function escapeClass( $class ) {
1168                 // Convert ugly stuff to underscores and kill underscores in ugly places
1169                 return rtrim( preg_replace(
1170                         array( '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ),
1171                         '_',
1172                         $class ), '_' );
1173         }
1174
1175         /**
1176          * Given HTML input, escape with htmlspecialchars but un-escape entities.
1177          * This allows (generally harmless) entities like &#160; to survive.
1178          *
1179          * @param string $html HTML to escape
1180          * @return string Escaped input
1181          */
1182         static function escapeHtmlAllowEntities( $html ) {
1183                 $html = Sanitizer::decodeCharReferences( $html );
1184                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
1185                 # hurt.
1186                 $html = htmlspecialchars( $html, ENT_QUOTES );
1187                 return $html;
1188         }
1189
1190         /**
1191          * Regex replace callback for armoring links against further processing.
1192          * @param array $matches
1193          * @return string
1194          */
1195         private static function armorLinksCallback( $matches ) {
1196                 return str_replace( ':', '&#58;', $matches[1] );
1197         }
1198
1199         /**
1200          * Return an associative array of attribute names and values from
1201          * a partial tag string. Attribute names are forces to lowercase,
1202          * character references are decoded to UTF-8 text.
1203          *
1204          * @param string $text
1205          * @return array
1206          */
1207         public static function decodeTagAttributes( $text ) {
1208                 if ( trim( $text ) == '' ) {
1209                         return array();
1210                 }
1211
1212                 $attribs = array();
1213                 $pairs = array();
1214                 if ( !preg_match_all(
1215                         self::getAttribsRegex(),
1216                         $text,
1217                         $pairs,
1218                         PREG_SET_ORDER ) ) {
1219                         return $attribs;
1220                 }
1221
1222                 foreach ( $pairs as $set ) {
1223                         $attribute = strtolower( $set[1] );
1224                         $value = Sanitizer::getTagAttributeCallback( $set );
1225
1226                         // Normalize whitespace
1227                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1228                         $value = trim( $value );
1229
1230                         // Decode character references
1231                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1232                 }
1233                 return $attribs;
1234         }
1235
1236         /**
1237          * Build a partial tag string from an associative array of attribute
1238          * names and values as returned by decodeTagAttributes.
1239          *
1240          * @param array $assoc_array
1241          * @return string
1242          */
1243         public static function safeEncodeTagAttributes( $assoc_array ) {
1244                 $attribs = array();
1245                 foreach ( $assoc_array as $attribute => $value ) {
1246                         $encAttribute = htmlspecialchars( $attribute );
1247                         $encValue = Sanitizer::safeEncodeAttribute( $value );
1248
1249                         $attribs[] = "$encAttribute=\"$encValue\"";
1250                 }
1251                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1252         }
1253
1254         /**
1255          * Pick the appropriate attribute value from a match set from the
1256          * attribs regex matches.
1257          *
1258          * @param array $set
1259          * @throws MWException When tag conditions are not met.
1260          * @return string
1261          */
1262         private static function getTagAttributeCallback( $set ) {
1263                 if ( isset( $set[5] ) ) {
1264                         # No quotes.
1265                         return $set[5];
1266                 } elseif ( isset( $set[4] ) ) {
1267                         # Single-quoted
1268                         return $set[4];
1269                 } elseif ( isset( $set[3] ) ) {
1270                         # Double-quoted
1271                         return $set[3];
1272                 } elseif ( !isset( $set[2] ) ) {
1273                         # In XHTML, attributes must have a value so return an empty string.
1274                         # See "Empty attribute syntax",
1275                         # http://www.w3.org/TR/html5/syntax.html#syntax-attribute-name
1276                         return "";
1277                 } else {
1278                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1279                 }
1280         }
1281
1282         /**
1283          * @param string $text
1284          * @return string
1285          */
1286         private static function normalizeWhitespace( $text ) {
1287                 return preg_replace(
1288                         '/\r\n|[\x20\x0d\x0a\x09]/',
1289                         ' ',
1290                         $text );
1291         }
1292
1293         /**
1294          * Normalizes whitespace in a section name, such as might be returned
1295          * by Parser::stripSectionName(), for use in the id's that are used for
1296          * section links.
1297          *
1298          * @param string $section
1299          * @return string
1300          */
1301         static function normalizeSectionNameWhitespace( $section ) {
1302                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1303         }
1304
1305         /**
1306          * Ensure that any entities and character references are legal
1307          * for XML and XHTML specifically. Any stray bits will be
1308          * &amp;-escaped to result in a valid text fragment.
1309          *
1310          * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1311          *   numericized (this way we're well-formed even without a DTD)
1312          * b. any numeric char refs must be legal chars, not invalid or forbidden
1313          * c. use lower cased "&#x", not "&#X"
1314          * d. fix or reject non-valid attributes
1315          *
1316          * @param string $text
1317          * @return string
1318          * @private
1319          */
1320         static function normalizeCharReferences( $text ) {
1321                 return preg_replace_callback(
1322                         self::CHAR_REFS_REGEX,
1323                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1324                         $text );
1325         }
1326
1327         /**
1328          * @param string $matches
1329          * @return string
1330          */
1331         static function normalizeCharReferencesCallback( $matches ) {
1332                 $ret = null;
1333                 if ( $matches[1] != '' ) {
1334                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1335                 } elseif ( $matches[2] != '' ) {
1336                         $ret = Sanitizer::decCharReference( $matches[2] );
1337                 } elseif ( $matches[3] != '' ) {
1338                         $ret = Sanitizer::hexCharReference( $matches[3] );
1339                 }
1340                 if ( is_null( $ret ) ) {
1341                         return htmlspecialchars( $matches[0] );
1342                 } else {
1343                         return $ret;
1344                 }
1345         }
1346
1347         /**
1348          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1349          * return the equivalent numeric entity reference (except for the core &lt;
1350          * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1351          * the HTML equivalent. Otherwise, returns HTML-escaped text of
1352          * pseudo-entity source (eg &amp;foo;)
1353          *
1354          * @param string $name
1355          * @return string
1356          */
1357         static function normalizeEntity( $name ) {
1358                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1359                         return '&' . self::$htmlEntityAliases[$name] . ';';
1360                 } elseif ( in_array( $name, array( 'lt', 'gt', 'amp', 'quot' ) ) ) {
1361                         return "&$name;";
1362                 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1363                         return '&#' . self::$htmlEntities[$name] . ';';
1364                 } else {
1365                         return "&amp;$name;";
1366                 }
1367         }
1368
1369         /**
1370          * @param int $codepoint
1371          * @return null|string
1372          */
1373         static function decCharReference( $codepoint ) {
1374                 $point = intval( $codepoint );
1375                 if ( Sanitizer::validateCodepoint( $point ) ) {
1376                         return sprintf( '&#%d;', $point );
1377                 } else {
1378                         return null;
1379                 }
1380         }
1381
1382         /**
1383          * @param int $codepoint
1384          * @return null|string
1385          */
1386         static function hexCharReference( $codepoint ) {
1387                 $point = hexdec( $codepoint );
1388                 if ( Sanitizer::validateCodepoint( $point ) ) {
1389                         return sprintf( '&#x%x;', $point );
1390                 } else {
1391                         return null;
1392                 }
1393         }
1394
1395         /**
1396          * Returns true if a given Unicode codepoint is a valid character in XML.
1397          * @param int $codepoint
1398          * @return bool
1399          */
1400         private static function validateCodepoint( $codepoint ) {
1401                 return $codepoint == 0x09
1402                         || $codepoint == 0x0a
1403                         || $codepoint == 0x0d
1404                         || ( $codepoint >= 0x20 && $codepoint <= 0xd7ff )
1405                         || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1406                         || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1407         }
1408
1409         /**
1410          * Decode any character references, numeric or named entities,
1411          * in the text and return a UTF-8 string.
1412          *
1413          * @param string $text
1414          * @return string
1415          */
1416         public static function decodeCharReferences( $text ) {
1417                 return preg_replace_callback(
1418                         self::CHAR_REFS_REGEX,
1419                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1420                         $text );
1421         }
1422
1423         /**
1424          * Decode any character references, numeric or named entities,
1425          * in the next and normalize the resulting string. (bug 14952)
1426          *
1427          * This is useful for page titles, not for text to be displayed,
1428          * MediaWiki allows HTML entities to escape normalization as a feature.
1429          *
1430          * @param string $text Already normalized, containing entities
1431          * @return string Still normalized, without entities
1432          */
1433         public static function decodeCharReferencesAndNormalize( $text ) {
1434                 global $wgContLang;
1435                 $text = preg_replace_callback(
1436                         self::CHAR_REFS_REGEX,
1437                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1438                         $text, /* limit */ -1, $count );
1439
1440                 if ( $count ) {
1441                         return $wgContLang->normalize( $text );
1442                 } else {
1443                         return $text;
1444                 }
1445         }
1446
1447         /**
1448          * @param string $matches
1449          * @return string
1450          */
1451         static function decodeCharReferencesCallback( $matches ) {
1452                 if ( $matches[1] != '' ) {
1453                         return Sanitizer::decodeEntity( $matches[1] );
1454                 } elseif ( $matches[2] != '' ) {
1455                         return Sanitizer::decodeChar( intval( $matches[2] ) );
1456                 } elseif ( $matches[3] != '' ) {
1457                         return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1458                 }
1459                 # Last case should be an ampersand by itself
1460                 return $matches[0];
1461         }
1462
1463         /**
1464          * Return UTF-8 string for a codepoint if that is a valid
1465          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1466          * @param int $codepoint
1467          * @return string
1468          * @private
1469          */
1470         static function decodeChar( $codepoint ) {
1471                 if ( Sanitizer::validateCodepoint( $codepoint ) ) {
1472                         return UtfNormal\Utils::codepointToUtf8( $codepoint );
1473                 } else {
1474                         return UtfNormal\Constants::UTF8_REPLACEMENT;
1475                 }
1476         }
1477
1478         /**
1479          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1480          * return the UTF-8 encoding of that character. Otherwise, returns
1481          * pseudo-entity source (eg "&foo;")
1482          *
1483          * @param string $name
1484          * @return string
1485          */
1486         static function decodeEntity( $name ) {
1487                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1488                         $name = self::$htmlEntityAliases[$name];
1489                 }
1490                 if ( isset( self::$htmlEntities[$name] ) ) {
1491                         return UtfNormal\Utils::codepointToUtf8( self::$htmlEntities[$name] );
1492                 } else {
1493                         return "&$name;";
1494                 }
1495         }
1496
1497         /**
1498          * Fetch the whitelist of acceptable attributes for a given element name.
1499          *
1500          * @param string $element
1501          * @return array
1502          */
1503         static function attributeWhitelist( $element ) {
1504                 $list = Sanitizer::setupAttributeWhitelist();
1505                 return isset( $list[$element] )
1506                         ? $list[$element]
1507                         : array();
1508         }
1509
1510         /**
1511          * Foreach array key (an allowed HTML element), return an array
1512          * of allowed attributes
1513          * @return array
1514          */
1515         static function setupAttributeWhitelist() {
1516                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
1517                 static $whitelist, $staticInitialised;
1518
1519                 $globalContext = implode( '-', compact( 'wgAllowRdfaAttributes', 'wgAllowMicrodataAttributes' ) );
1520
1521                 if ( $whitelist !== null && $staticInitialised == $globalContext ) {
1522                         return $whitelist;
1523                 }
1524
1525                 $common = array(
1526                         # HTML
1527                         'id',
1528                         'class',
1529                         'style',
1530                         'lang',
1531                         'dir',
1532                         'title',
1533
1534                         # WAI-ARIA
1535                         'role',
1536                 );
1537
1538                 if ( $wgAllowRdfaAttributes ) {
1539                         # RDFa attributes as specified in section 9 of
1540                         # http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1541                         $common = array_merge( $common, array(
1542                                 'about', 'property', 'resource', 'datatype', 'typeof',
1543                         ) );
1544                 }
1545
1546                 if ( $wgAllowMicrodataAttributes ) {
1547                         # add HTML5 microdata tags as specified by
1548                         # http://www.whatwg.org/html/microdata.html#the-microdata-model
1549                         $common = array_merge( $common, array(
1550                                 'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1551                         ) );
1552                 }
1553
1554                 $block = array_merge( $common, array( 'align' ) );
1555                 $tablealign = array( 'align', 'valign' );
1556                 $tablecell = array(
1557                         'abbr',
1558                         'axis',
1559                         'headers',
1560                         'scope',
1561                         'rowspan',
1562                         'colspan',
1563                         'nowrap', # deprecated
1564                         'width', # deprecated
1565                         'height', # deprecated
1566                         'bgcolor', # deprecated
1567                 );
1568
1569                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1570                 # See: http://www.w3.org/TR/html4/
1571                 $whitelist = array(
1572                         # 7.5.4
1573                         'div'        => $block,
1574                         'center'     => $common, # deprecated
1575                         'span'       => $common,
1576
1577                         # 7.5.5
1578                         'h1'         => $block,
1579                         'h2'         => $block,
1580                         'h3'         => $block,
1581                         'h4'         => $block,
1582                         'h5'         => $block,
1583                         'h6'         => $block,
1584
1585                         # 7.5.6
1586                         # address
1587
1588                         # 8.2.4
1589                         'bdo'        => $common,
1590
1591                         # 9.2.1
1592                         'em'         => $common,
1593                         'strong'     => $common,
1594                         'cite'       => $common,
1595                         'dfn'        => $common,
1596                         'code'       => $common,
1597                         'samp'       => $common,
1598                         'kbd'        => $common,
1599                         'var'        => $common,
1600                         'abbr'       => $common,
1601                         # acronym
1602
1603                         # 9.2.2
1604                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1605                         'q'          => array_merge( $common, array( 'cite' ) ),
1606
1607                         # 9.2.3
1608                         'sub'        => $common,
1609                         'sup'        => $common,
1610
1611                         # 9.3.1
1612                         'p'          => $block,
1613
1614                         # 9.3.2
1615                         'br'         => array_merge( $common, array( 'clear' ) ),
1616
1617                         # http://www.whatwg.org/html/text-level-semantics.html#the-wbr-element
1618                         'wbr'        => $common,
1619
1620                         # 9.3.4
1621                         'pre'        => array_merge( $common, array( 'width' ) ),
1622
1623                         # 9.4
1624                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1625                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1626
1627                         # 10.2
1628                         'ul'         => array_merge( $common, array( 'type' ) ),
1629                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1630                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1631
1632                         # 10.3
1633                         'dl'         => $common,
1634                         'dd'         => $common,
1635                         'dt'         => $common,
1636
1637                         # 11.2.1
1638                         'table'      => array_merge( $common,
1639                                                                 array( 'summary', 'width', 'border', 'frame',
1640                                                                                 'rules', 'cellspacing', 'cellpadding',
1641                                                                                 'align', 'bgcolor',
1642                                                                 ) ),
1643
1644                         # 11.2.2
1645                         'caption'    => $block,
1646
1647                         # 11.2.3
1648                         'thead'      => $common,
1649                         'tfoot'      => $common,
1650                         'tbody'      => $common,
1651
1652                         # 11.2.4
1653                         'colgroup'   => array_merge( $common, array( 'span' ) ),
1654                         'col'        => array_merge( $common, array( 'span' ) ),
1655
1656                         # 11.2.5
1657                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1658
1659                         # 11.2.6
1660                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1661                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1662
1663                         # 12.2
1664                         # NOTE: <a> is not allowed directly, but the attrib
1665                         # whitelist is used from the Parser object
1666                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1667
1668                         # 13.2
1669                         # Not usually allowed, but may be used for extension-style hooks
1670                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1671                         # true
1672                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1673
1674                         # 15.2.1
1675                         'tt'         => $common,
1676                         'b'          => $common,
1677                         'i'          => $common,
1678                         'big'        => $common,
1679                         'small'      => $common,
1680                         'strike'     => $common,
1681                         's'          => $common,
1682                         'u'          => $common,
1683
1684                         # 15.2.2
1685                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1686                         # basefont
1687
1688                         # 15.3
1689                         'hr'         => array_merge( $common, array( 'width' ) ),
1690
1691                         # HTML Ruby annotation text module, simple ruby only.
1692                         # http://www.whatwg.org/html/text-level-semantics.html#the-ruby-element
1693                         'ruby'       => $common,
1694                         # rbc
1695                         'rb'         => $common,
1696                         'rp'         => $common,
1697                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1698                         'rtc'         => $common,
1699
1700                         # MathML root element, where used for extensions
1701                         # 'title' may not be 100% valid here; it's XHTML
1702                         # http://www.w3.org/TR/REC-MathML/
1703                         'math'       => array( 'class', 'style', 'id', 'title' ),
1704
1705                         # HTML 5 section 4.6
1706                         'bdi' => $common,
1707
1708                         # HTML5 elements, defined by:
1709                         # http://www.whatwg.org/html/
1710                         'data' => array_merge( $common, array( 'value' ) ),
1711                         'time' => array_merge( $common, array( 'datetime' ) ),
1712                         'mark' => $common,
1713
1714                         // meta and link are only permitted by removeHTMLtags when Microdata
1715                         // is enabled so we don't bother adding a conditional to hide these
1716                         // Also meta and link are only valid in WikiText as Microdata elements
1717                         // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1718                         // So we don't bother including $common attributes that have no purpose.
1719                         'meta' => array( 'itemprop', 'content' ),
1720                         'link' => array( 'itemprop', 'href' ),
1721                 );
1722
1723                 $staticInitialised = $globalContext;
1724
1725                 return $whitelist;
1726         }
1727
1728         /**
1729          * Take a fragment of (potentially invalid) HTML and return
1730          * a version with any tags removed, encoded as plain text.
1731          *
1732          * Warning: this return value must be further escaped for literal
1733          * inclusion in HTML output as of 1.10!
1734          *
1735          * @param string $text HTML fragment
1736          * @return string
1737          */
1738         static function stripAllTags( $text ) {
1739                 # Actual <tags>
1740                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1741
1742                 # Normalize &entities and whitespace
1743                 $text = self::decodeCharReferences( $text );
1744                 $text = self::normalizeWhitespace( $text );
1745
1746                 return $text;
1747         }
1748
1749         /**
1750          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1751          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1752          * PHP 5.1 doesn't.
1753          *
1754          * Use for passing XHTML fragments to PHP's XML parsing functions
1755          *
1756          * @return string
1757          */
1758         static function hackDocType() {
1759                 $out = "<!DOCTYPE html [\n";
1760                 foreach ( self::$htmlEntities as $entity => $codepoint ) {
1761                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1762                 }
1763                 $out .= "]>\n";
1764                 return $out;
1765         }
1766
1767         /**
1768          * @param string $url
1769          * @return mixed|string
1770          */
1771         static function cleanUrl( $url ) {
1772                 # Normalize any HTML entities in input. They will be
1773                 # re-escaped by makeExternalLink().
1774                 $url = Sanitizer::decodeCharReferences( $url );
1775
1776                 # Escape any control characters introduced by the above step
1777                 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
1778                         array( __CLASS__, 'cleanUrlCallback' ), $url );
1779
1780                 # Validate hostname portion
1781                 $matches = array();
1782                 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1783                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1784
1785                         // Characters that will be ignored in IDNs.
1786                         // http://tools.ietf.org/html/3454#section-3.1
1787                         // Strip them before further processing so blacklists and such work.
1788                         $strip = "/
1789                                 \\s|          # general whitespace
1790                                 \xc2\xad|     # 00ad SOFT HYPHEN
1791                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1792                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1793                                 \xe2\x81\xa0| # 2060 WORD JOINER
1794                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1795                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1796                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1797                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1798                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1799                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1800                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1801                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
1802                                 /xuD";
1803
1804                         $host = preg_replace( $strip, '', $host );
1805
1806                         // @todo FIXME: Validate hostnames here
1807
1808                         return $protocol . $host . $rest;
1809                 } else {
1810                         return $url;
1811                 }
1812         }
1813
1814         /**
1815          * @param array $matches
1816          * @return string
1817          */
1818         static function cleanUrlCallback( $matches ) {
1819                 return urlencode( $matches[0] );
1820         }
1821
1822         /**
1823          * Does a string look like an e-mail address?
1824          *
1825          * This validates an email address using an HTML5 specification found at:
1826          * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
1827          * Which as of 2011-01-24 says:
1828          *
1829          *   A valid e-mail address is a string that matches the ABNF production
1830          *   1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
1831          *   in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
1832          *   3.5.
1833          *
1834          * This function is an implementation of the specification as requested in
1835          * bug 22449.
1836          *
1837          * Client-side forms will use the same standard validation rules via JS or
1838          * HTML 5 validation; additional restrictions can be enforced server-side
1839          * by extensions via the 'isValidEmailAddr' hook.
1840          *
1841          * Note that this validation doesn't 100% match RFC 2822, but is believed
1842          * to be liberal enough for wide use. Some invalid addresses will still
1843          * pass validation here.
1844          *
1845          * @since 1.18
1846          *
1847          * @param string $addr E-mail address
1848          * @return bool
1849          */
1850         public static function validateEmail( $addr ) {
1851                 $result = null;
1852                 if ( !Hooks::run( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
1853                         return $result;
1854                 }
1855
1856                 // Please note strings below are enclosed in brackets [], this make the
1857                 // hyphen "-" a range indicator. Hence it is double backslashed below.
1858                 // See bug 26948
1859                 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
1860                 $rfc1034_ldh_str = "a-z0-9\\-";
1861
1862                 $html5_email_regexp = "/
1863                 ^                      # start of string
1864                 [$rfc5322_atext\\.]+    # user part which is liberal :p
1865                 @                      # 'apostrophe'
1866                 [$rfc1034_ldh_str]+       # First domain part
1867                 (\\.[$rfc1034_ldh_str]+)*  # Following part prefixed with a dot
1868                 $                      # End of string
1869                 /ix"; // case Insensitive, eXtended
1870
1871                 return (bool)preg_match( $html5_email_regexp, $addr );
1872         }
1873 }